The formation of GaAs(110) interfaces with Ni was investigated with soft x-ray photoelectron spectroscopy at the Super-ACO synchrotron radiation source at L.U.R.E. in Orsay (France). Ni was deposited on p-GaAs(110) surfaces cleaved in ultrahigh vacuum (UHV), either by direct evaporation at substrate temperatures of 300 K [room temperature (RT)] or 20 K [low temperature (LT)] or by evaporation on top of an Ar layer condensed at 20 K on the semiconductor surface. This Ar layer was further desorbed by annealing around 100 K, and the Ni could thus be brought smoothly onto the GaAs(110) surface. This method allows the formation of nonreacted interfaces with a metal which, when deposited directly at RT or LT, brings a strong degree of surface disruption. Ga 3d and As 3d core level spectra were recorded on surfaces with overlayers from 0.01 to 10 Å, together with the valence band (VB) spectra in normal conditions (photon energy hν=50 eV) and in resonant conditions (hν=67 eV, corresponding to the photoemission threshold for the Ni 3p core levels). A reaction component was observed on the Ga 3d spectra for Ni thicknesses as small as 0.2 Å on the direct interfaces formed at RT as well as LT (direct interface meaning interface formed by direct evaporation). No reaction with the GaAs surface could be detected when the Ni is deposited on the Ar buffer layer, even for a thickness of 15 Å. The valence band spectra indicate that NiAs is formed for direct interfaces. At LT, this reaction is limited to the first atomic layer, but it is observed to extend up to 10 Å thick layers for deposition at RT. The evolution of the intensity of the core level and VB spectra clearly indicates the formation of NiAs clusters at RT. The evolution of the Schottky barrier height differs for the two types of deposition. For the direct evaporation, Ef saturates [on p-type GaAs(110)] at 500 meV above the valence band maximum (VBM), the kinetics being slower at LT due to photoelectron perturbation of the band bending (photovoltage effect); for deposition on the Ar buffer layer, both the final position (300 meV above VBM) and the kinetics are different.